European Review for Medical and Pharmacological Sciences 2019; 23: 7532-7542 Comparison of Billroth I, Billroth II, and Roux-en-Y reconstructions after distal gastrectomy according to functional recovery: a meta-analysis X.- F. LIU1, Z.-M. GAO1, R.-Y. WANG2, P.-L. WANG1, K. LI1, S. GAO3 1Department of Surgical Oncology and General Surgery, the First Affiliated Hospital of China Medical University, Shenyang, China 2Department of Ultrasound, the First Affiliated Hospital of China Medical University, Shenyang, China 3Department of Gynecology and Obstetrics, the ShengJing Hospital of China Medical University, Shenyang, China Xiao-Fang Liu and Zi-Ming Gao have to be considered as first authors Abstract. – OBJECTIVE: Gastric cancer is Introduction common, with a high mortality rate. Billroth I (B- I), Billroth II (B-II), and Roux-en-Y (R-Y) are the ma- Gastric cancer is the fourth most common jor reconstruction procedures after distal gastrec- malignancy and the second leading cause of tomy. In our study, we aimed to evaluate the func- tional recovery following the B-I, B-II, and R-Y re- cancer-related death worldwide, with about constructions through a network meta-analysis. 951,600 new cases and 723,100 deaths every MATERIALS AND METHODS: PubMed, Em- year1,2. Up to now, surgical resection is still base, and Cochrane Library databases were the most effective measure for gastric cancer, searched until April 2018. From the included especially for the early stage. Distal gastrecto- studies, first oral-intake time, early complica- my is recommended for the mid-lower gastric tions, endoscopic finding, quality of life (QoL), and body weight changes were extracted as tumors, which account for approximately 70% the short- and long-term outcomes of recon- of gastric tumors and always have a better structions. The network meta-analysis was per- prognosis than the upper tumors. formed with R 3.4.2 software as well as “gemtc” Billroth I (B-I), Billroth II (B-II), and Roux- and “forestplot” packages. RESULTS: en-Y (R-Y) are the major reconstruction proce- Our work included a total of 26 arti- dures after distal gastrectomy. Also, some of cles involving 6212 patients with gastric cancer. Network meta-analysis revealed that R-Y recon- their modified procedures have been introduced struction has a lower risk and degree of resid- in the last decade. However, the most appropri- ual gastritis and bile reflex than B-I and B-II re- ate reconstruction procedure among the three constructions. However, no differences in first still remains unclear for surgeons. For early oral-intake time, complications, risk of reflux gastric cancer, B-I is a common reconstruction esophagitis, and residual food, QoL, and body weight changes existed among the three recon- technology, especially in Japan and Korea. structions. However, B-II and R-Y are mostly performed in CONCLUSIONS: R-Y may be the appropriate advanced western countries. Indeed, efforts to reconstruction procedure after distal gastrec- compare these three reconstructions have been tomy based on postoperative functional recov- made in some systematic reviews3,4. However, ery. However, more reports with a large sample due to the lack of reasonable methods, their size are warranted to investigate its long-term outcomes. effective evaluation was insufficient. Thus, this work aimed to evaluate the functional recovery Key Words: following the B-I, B-II, and R-Y reconstruc- Gastric cancer, Distal gastrectomy, Reconstruction, tions through a Bayesian network meta-analy- Functional recovery. sis with relatively comprehensive data. Corresponding Authors: Shan Gao, MD; e-mail: [email protected] 7532 Kai Li, MD; e-mail: [email protected] B-I, B-II, R-Y comparison after distal gastrectomy: a meta-analysis Material and Methods as the short-term outcomes. Moreover, the endo- scopic finding, QoL, and body weight changes Literature Searches and Study Selection were evaluated as the long-term outcomes. A systematic search of PubMed, Embase, and Cochrane Library was performed until April Statistical Analysis 2018. The search terms included “(Stomach Neo- We performed the network meta-analysis with plasms) OR gastric cancer) OR gastric carcino- R software and “gemtc” and “forest plot” pack- ma”, “(Distal) OR Subtotal”, “Gastrectomy”, and ages. We established the random-effects network “(Reconstructive Surgical Procedures) OR Re- models, fitting a generalized linear model under constructive Surgery) OR reconstruction”. Only the Bayesian framework, by linking to JAGS. English articles were included in our analysis. Odds ratios (ORs) and mean difference (MD) Two reviewers screened the search results inde- with corresponding 95% confidential intervals pendently. The procedures of this meta-analysis (CIs) were calculated for binary and continuous followed the PRISMA guidelines. outcomes, respectively. We also estimated the consistency between direct and indirect evidence Inclusion and Exclusion Criteria using a node-splitting method. Moreover, if the The inclusion criteria were as follows: (1) results of node-splitting were p>0.05, the consis- gastric cancer diagnosed clearly according to tency model was selected for the meta-analysis. pathological findings; (2) distal gastrectomy, re- All analyses were performed with R 3.4.2 soft- gardless of open or laparoscopy, performed for ware (R Foundation for Statistical Computing, each patient; (3) B-I, B-II, or R-Y, including the Vienna, Austria). modified procedures, chosen as the digestive re- construction procedure after distal gastrectomy; (4) no prior chemotherapy or radiation therapy Results before gastrectomy. However, studies with dupli- cated data, letters, comments, reviews, abstracts, Baseline Characteristics of Studies or editorial articles were excluded. A total of 964 potential researches without duplication were searched through PubMed, Em- Data Extraction and Quality Assessment base, and Cochrane Library databases. According The following data were extracted from each to the criteria, 26 studies with 6212 patients were study by two reviewers: first author, year of pub- finally included in our work5-30, of which 9 and 17 lication, study design, age, proportion of male, were RCTs and case-control studies, respectively. TNM stage, reconstruction techniques, first The literature screening process is shown in Fig- oral-intake time, early complications (including ure 1, the characteristics of the included studies Clavien-Dindo Classification), RGB Classifica- are listed in Table I, and the quality of literature tion of residual stomach, Los Angeles Classifica- is presented in S1 Table. tion of lower esophagus, quality of life (QoL) of EORTC QLQ-C30, and study population sizes. For analysis of bias risks, two reviewers assessed Short-term Outcomes the quality of literature. The modified Jadad scale and Newcastle-Ottawa Quality Assessment Scale First Oral-Intake Time (NOS) were used for randomized controlled trial To compare the postoperative recovery of gas- (RCT) and cohort or case-control studies, respec- trointestinal function, 11 studies6,9-12,14,18,20,25,27,30 tively. reporting the first oral-intake time were included. As shown in Figure 2a, there was no difference Evaluation of Short- and in short-term gastrointestinal function restoration Long-term Outcomes between B-I, B-II, and R-Y reconstructions. Given that no significant difference in progno- sis existed between B-I, B-II, and R-Y, we mainly Complications compared the postoperative functional recovery, A total of 16 studies7,10-12,14,17-19,22-29 were included which is being strongly considered by surgeons to compare the overall complications. The over- nowadays. We regarded the first oral-intake time all complication rates of B-I, B-II, and R-Y were and postoperative complications, which could 11.6%, 20.2%, and 19.8%, respectively. However, reflect the early recovery of gastrointestinal tract, according to our network meta-analysis results, no 7533 X.-F. Liu, Z.-M. Gao, R.-Y. Wang, P.-L. Wang, K. Li, S. Gao Figure 1. Flowchart of the study search and selection process. difference existed between B-I, B-II, and R-Y (Fig- included. Of these, 8 studies assessed reflux ure 2b). In addition, another 10 studies9,14,18,20,23-27,30, esophagitis according to the Los Angeles Classi- which adopted the Clavien-Dindo Classification, fication and 14 researches5,7,8,12,16,18-20,22,23,25,27,28,30 were analyzed for the severity of complications. assessed gastric residue, gastritis, and bile re- Similar to the overall result, there was no difference flux according to the RGB classification. No in the incidence rates of grade III-IV complications difference existed between B-I, B-II, and R-Y in between these three reconstructions (Figure 2b). terms of risk of reflux esophagitis and residual food. However, the gastritis and bile reflex were relatively more unusual for R-Y group than for Long-term Outcomes B-I and B-II groups [R-Y vs. B-II: OR (95% CIs)=0.08 (0.03, 0.20); R-Y vs. B-I: OR (95% Endoscopic Finding CIs)=0.17 (0.075, 0.38)] (Figure 3a). Moreover, A total of 17 reports7,15,16,20,22,23,30 that com- as shown in Figure 3b, we evaluated the severity pared the difference in endoscopic findings were of endoscopic findings using the RGB classifi- 7534 B-I, B-II, R-Y comparison after distal gastrectomy: a meta-analysis Table I. Original characteristics of included studies. Open/ Gender TNM ID Study Design Laparoscopy Reconstruction No. Male/Female Age stage Outcomes 1 2002 Retrospective Open B-I 175 None None None EF